This post is merely for people who get a specific error message and has nothing to do with anything else. It's meant to be cataloged by the search engines to help people. This is not a normal blog posting.

Recently I tried running my Adobe Premiere video editing package and it wouldn't work. I received the following error:

Cannot load transcoding service
Xerces Panic Error

I spent 3 hours debugging this problem - I followed message board suggestions, deleted registry key trees (bad idea!) and paid for a registry cleaning utility that did "something" but didn't address my Xerces issue.

This error, from searching, happens with a lot of video and still editing packages - Adobe Premiere Pro, PDF Creator, Photoshop some other still imaging products made by other companies, ArcIMS Monitor service, Dreamweaver, etc.

The problem is not specifically Adobe, or other manufacturers, but with the darned included Xerces product. One site suggested removing the entire registry tree for \HKEY_CLASSES_ROOT\MIME\DataBase\Charset\ - let me pose a question - if the tree were removable, would it not be in the registry in the first place?

I finally found a good solution over at IBM's support page. I copy/paste the details here in case it ever goes away:

'Xerces Panic Error' when launching Viewer
Technote (FAQ)

Problem
When you launch the Viewer, a message is displayed reading "Xerces Panic Error."

Solution
The Viewer uses Xerces to read the XML code that the PureEdge forms are based on. The Xerces parser is trying to access the charset keys within the registry, and the user may not have access to read them. The keys of significance can be located at:

HKEY_CLASSES_ROOT\MIME\DataBase\Charset\

To open the registry and view the keys, launch the Registry Editor by selecting Start -> Run -> and type "regedit".

In the folder view on the left, navigate to \HKEY_CLASSES_ROOT\MIME\DataBase\Charset\

The user must be given read access to those keys.

If the user already has access to the keys, the problem may be that one of the keys is not as it should be.

Every key within the charset directory must contain one (and only one, never both) of the following:
1. AliasForCharset
2. CodePage
- if the key has a CodePage element then it must also have an "InternetCoding" element

If either of the elements are mispelled, they need to be modified.

So what I did was export the branch of the tree to a text editor for faster searching. I opened it in my favorite text editor (UltraEdit) and scanned it fairly quickly. I found one of the entries has no sub-key. Note as the IBM tech document states, each charset directory my contain exactly one of AliasForCharset or CodePage. I had one that had neither, removed that key, and everything worked fine afterward.

The real question is why is Xerces so darned particular. If there's nothing there, just go on and ignore the key please... or at least report back specifically what the problem was. Idiot coders...

Good luck!